Labour Cost and What It Hides

Labour as a percentage of sales is the metric that drives most scheduling decisions. It is useful and it is incomplete in ways that matter.

What the percentage contains

Wages for hours scheduled and worked.

Sometimes on-costs; frequently not.

Measured weekly, by site, against sales.

Which makes it comparable, actionable and the thing a site manager is judged on.

What it does not contain

Turnover. Replacing a trained person in these sectors costs the equivalent of many weeks of the margin being protected, and it appears in a recruitment line rather than in labour cost.

Predictability pay, where it applies, which is frequently booked centrally.

Overtime caused by under-scheduling, which lands in the same week and is sometimes attributed elsewhere.

Lost sales from under-staffing, which is invisible by construction: nobody counts the customer who left the queue.

The incentive this creates

A site manager cuts hours to hit the percentage.

Service degrades, the queue lengthens, some customers leave.

Sales fall, so the percentage stays roughly the same, and the cut looks neutral.

Meanwhile turnover rises, which is somebody else's number.

The metric is locally optimised and globally wrong, which is not the manager's fault.

Making the hidden costs visible

Book predictability pay to the site that caused it.

Report turnover by site alongside labour percentage, on the same page.

Track overtime hours separately from scheduled hours, because they are a planning signal and not a cost line.

And attempt some measure of service: queue length, wait time, abandoned baskets, service complaints — anything that moves when staffing is cut.

The under-staffing test

Compare the sites with the lowest labour percentage against the sites with the best service measures.

If they are different sites, the percentage is being achieved by cutting service.

That is a finding about the metric, and it is the argument for adding the second column rather than for admonishing anyone.

What to schedule to

The forecast plus a buffer, where the buffer is sized from measured forecast error rather than guessed.

Which costs a small, known amount and removes the cuts, the call-ins and the predictability pay that the precise version generates.

Run that comparison once and the buffer usually justifies itself.

Report turnover alongside it

On the same page, by site.

The percentage is locally optimised and globally wrong when service and retention are missing.

Adding one column changes what the number means without changing how it is calculated.

And the sites with the best percentage and the worst retention become visible immediately, which is the finding the metric alone conceals.

A practical implementation prompt

During configuration, use project time tracking application to prompt questions about fields, ownership and output. Keep the written scheduling purpose in control and document every assumption.